The devastating impacts of seismic events during the last decades in areas with densely concentrated population and buildings pointed out that these environments are highly exposed to human and economic losses. In risk analysis the probability of losses is calculated over a specified period of time due to all the possible future seis- mic events, whereas in a seismic scenario the impact of a given earthquake is investigated and quantified. Reli- able earthquake loss estimation (in monetary terms) for buildings struck by an earthquake is of growing impor- tance both for the planning of appropriate and cost effective earthquake mitigation measures and for insurance purposes, and also for the definition of criteria for prioritizing seismic strengthening (rehabilitation) programmes for existing buildings. Decisions regarding the seismic rehabilitation of existing buildings require both engi- neering and economic studies and consideration of social priorities. Pre- and post-earthquake upgrading of a city’s existing building stock is one of the most conflictual and difficult issues of public policy decisions.

The interest in earthquake management by governments and policies is obvious considering the numerous projects financed for this purpose. After the socio-economic impact of the earthquakes in Turkey (Izmit on 17^th-8-1999 & Düzce on 12^th-11-1999), and Greece (Athens on 7^th-9-1999), the European Commission funded in 1999 the RISK-UE project: “An advanced approach to earthquake risk scenarios with application to different European towns”, aiming at the assessment of seismic risk in European urban centres. Seven research centres from European countries (France, Italy, Romania, Spain, Greece, FYROM, and Bulgaria) were involved in the project. Shortly before RISK-UE, another international project, RADIUS (Risk Assessment Tools for Diagnosis of Urban Areas against Seismic Disasters) aimed to develop earthquake damage scenarios in urban areas of nine case-study cities all over the world. The ENSeRVES project (European Network on Seismic Risk, Vulnerability and Earthquake Scenarios) in 1997 gathered together teams of scientists of different categories (seismologists, geologists, engineers, architects, etc.) involving 11 international institutions working on earthquake engineering and seismology. Along similar lines, the National Technical Chamber of Greece (NTCG) with the cooperation of Greek universities provided in 1996 funding to the Earthquake Planning and Protection Organization (EPPO) for carrying out the “National Programme for Earthquake Management of Existing Buildings” [1] . The project has been conducted by several regional sections of the NTCG and involved applications in selected Greek cities (Xanthi, 2005 & 2007; Tripoli, 2004; Corfu, 2005), with significant results.

The evaluation of loss due to building damage in an area struck by an earthquake depends both on seismic hazard and the vulnerability of the building inventory in the certain region. Loss is defined as the human and financial consequences of damage, including injuries or deaths or the costs of repair. A wider research of seis- mic risk assessment could include direct economic losses (Figure 1) of infrastructures (water networks, sewer- age, roads, bridges, etc.) and indirect losses (human losses, business interruption). The seismic vulnerability of a building can be defined as its proneness to be damaged by an earthquake. Seismic vulnerability relationships at- tempt to predict for several building classes the mean degree and the extent of damage at given levels of seismic demand. Vulnerability analysis reveals the damageability of the structure(s) under varying intensity or magni- tudes of ground motion. Multiple damage states are typically considered in the analysis [4] .

Based on a quantitative assessment of seismic vulnerability, the probability of damage to given building types caused by earthquakes of various intensities can be predicted [5] .

It is important to clarify the distinction between risk and vulnerability. Risk combines the expected losses from all levels of hazard severity, also taking their occurrence probability into account, while vulnerability of an element is usually expressed for a given hazard severity level. Components of seismic risk assessment and loss estimation are 1) Hazard analysis; 2) Local site effects (microzonation); 3) Exposure information (structural in- ventory); 4) Vulnerability analysis; 5) Estimation of risk and loss. Since the standard definition of risk is a pro- bability or likelihood of loss, between zero and one, it may be more appropriate to express risk as Risk = Hazard × Vulnerability while loss depends on the value of the exposure at risk, given by Loss = Hazard × Vulnerability × Exposure. Thus, while seismic hazard is a product of natural processes, seismic risk and loss are dependent on the vulnerability and social exposure in terms of the built environment, human population, and value of opera- tions.

The first step for the development of any earthquake scenario is the assessment of damage in structures. Sev- eral methodologies and relations exist attempting to express damage indices in economic loss. The correlation of structural damage to economic loss is indispensable for the estimation of seismic risk [6] -[8] . Many seismic risk assessments and vulnerability studies [9] -[19] have been carried out, and their results constitute important tools in the mitigation of losses due to future seismic events, e.g. allowing disaster management plans to be drawn up.

The research includes a study for the seismic vulnerability and risk assessment in the extended region of Ath- ens (Greece) struck by the 7^th-9-1999 Parnitha’s earthquake. The building stock in the study area consists of typical building types, representative of the materials, seismic codes and construction techniques of Southern Europe. The building exposure refers to 750,085 buildings which are situated in 122 regions of Attica according to the results of the 2000 statistical census (one year after the seismic event), information obtained from the Na- tional Statistics Service of Greece (NSSG). For the evaluation of seismic hazard, data specific to the characteris- tics of the earthquake that struck the area has been used. The seismic demand is characterized by the ratio, , where is the regional PGA which is evaluated using simple expressions from the estimated in ear- lier research macroseismic intensities, and is the PGA by which each municipality of Attica is characterized according to the hazard map of the 2003 Seismic Code [20] . A pilot methodology is developed for the seismic loss assessment in monetary terms, consistent with the National Programme for Earthquake Management of Ex- isting Buildings (NPEMEB). Useful results, which had been derived from the application of the specific project in several Greek cities, have been used for the needs of this study. The vulnerability assessment is based on four different existing damage scenarios. The three of them have been proposed by NTCG in 2006 and the vulner-

ability curves have been derived from a hybrid approach, which combines statistical data with appropriately processed results from nonlinear dynamic or static analyses, that permit extrapolation of statistical data to PGA’s and/or spectral displacements for which no data are available. The forth damage scenario is based on relatively recently developed Damage Probability Matrices-DPMs [21] applying the empirical seismic method for the vulnerability analysis on a large set of observational data comprising 180,945 buildings which developed damage of varying degree, type and extent after the 7^th of September 1999 Parnitha’s earthquake. The empirical vulnerability assessment is generally based on the distribution of damage reported in post-earthquake surveys and treats these data according to statistical procedures. It includes the real response of the exposed building stock, taking into account all the structural characteristics, topography, site and soil conditions of Greece. Sur- vey data can rarely provide a complete set of data. The difficulty focuses on the lack of a sufficiently large set of reliable empirical data, due to the limited number of damaging earthquakes at a small distance from densely populated areas, covering a wide range of ground motions [19] [20] .

Information regarding the compatible (budget approved according to the ministry’s provisions) repair cost af- ter the 1999 Parnitha’s earthquake has been used in order to conduct correlation analysis with the estimated losses. The statistically derived repair cost for the area is compared with the results of the economic loss estima- tion obtained using the pre-described procedure for the risk assessment. The comparison of the estimated eco- nomic loss with the compatible repair cost calibrates the reliability of the commonly used method for the risk assessment and serves in the improvement of seismic security and prioritizing the criteria for seismic rehabilita- tion programmes of existing buildings.

The development of seismic vulnerability and risk models needs a classification system to characterize the earthquake-exposed building stock and describe its damage. A complete set of data (i.e. covering the entire city) is able to be provided only by the National Statistics Service of Greece (NSSG). The current research is focused on the seismic risk assessment of Attica area struck by the 7^th-9-1999 Parnitha’s earthquake and refers to 750,085 buildings which are located in 122 regions. The above information has been derived from NSSG ac- cording to the information of 2000-1 census of buildings, conducting just a year after the occurrence of the earthquake. According to the same source, the building exposure in Attica represents the 18.8% (/3990970) of total population of the entire building stock in Greece (total number of buildings). A full set of data collected from NSSG regarding: 1) The total number of buildings of the study area (Attica); 2) The number of buildings categorized according to the construction materials (reinforced concrete, masonry, metal or wood or stone or other); 3) The number of buildings categorized according to the construction materials combined with the year of construction (Seismic Code); 4) The number of buildings categorized according to the construction materials and the period of construction combined with the height (number of floors). The classification system should also take into account the building types of the existing vulnerability models. The level of seismic design and construction detailing, could generally be discriminated in four subclasses, as follows: a) Without Seismic Code (or pre-seismic code: year of construction before 1959): RC buildings with practical very low level of seismic design or no seismic design, and poor quality of detailing; b) Low Seismic Code (the 1^st Greek Seismic Code of 1959: year of construction 1959-1985): RC buildings with low level of seismic design (corresponding approxi- mately to pre-1980 codes in Southern Europe); c) Moderate Seismic Code (the 1^st Greek Seismic Code of 1959 plus the 1985 Supplement Clauses: year of construction between 1985-1995): R/C buildings with medium level of seismic design (corresponding approximately to post-1980 codes in S. Europe) and reasonable seismic de- tailing of R/C members; d) High Seismic Code (new Greek generation of RC and seismic codes similar to Eurocodes: year of construction after 1995): R/C buildings with adequate level of seismic design according to the new generation of seismic codes and ductile seismic detailing of R/C members including sufficient descrip- tions for detailing and anchorage. Useful information about the building exposure of Attica, which represents a reliable sample of Greece and generally South Europe, after elaborating the initial data collected from NSSG. The analysis results are presented in Tables 1-3 including the building stock regarding construction materials, period of construction connected to the seismic codes and number of floors.

Information that was missing for the needs of the research was derived from previous studies (Xanthi, 2005; 2007; Tripoli, 2004; Corfu, 2005) of the National Programme for Earthquake Management of Existing Buildings. Thus, buildings of Attica with 3 to 5 floors are distributed as follows: 1) RC buildings designed and constructed

earlier than 1985: 47.0% with 3 floors, 38.6% with 4 floors and 14.4% with 5 floors; 2) RC buildings designed and constructed between 1985 - 1995: 39.1% with 3 floors, 27.6% with 4 floors and 33.3% with 5 floors; 3) RC buildings designed and constructed after 1995: 54.5% with 3 floors, 25.5% with 4 floors and 20.0% with 5 floors; 4) masonry buildings are considered with 3 floors. In the category of six and more floors all buildings are considered having six floors. As far as RC buildings of Attica with ground floor without infill panels (pilotis) are regarded: a) 24.9% buildings designed and constructed earlier than 1985; b) 57.9% buildings designed and constructed between the period 1985 - 1995; and c) 59.7% buildings after 1995. Finally, the distribution of the mean constructed area per floor based on the previous studies has occurred: a) 150 m ² for buildings of RC structural system designed and constructed earlier than 1985; b) 133 m ² for RC buildings designed and constructed be- tween the period 1985 - 1995; c) 180 m ² for buildings of reinforced concrete structural system after 1995; and d) 74 m ² for masonry buildings.

Four different damage scenarios according to existing vulnerability curves are considered for the seismic risk assessment. These vulnerability models (in form of curves or DPMs) regarding typical structural types have been proposed by National Technical Chamber of Greece in 2006 (7 structural building types in 3 different damage scenarios) and also by Eleftheriadou [21] on the recently developed Damage Probability Matrices (5 structural building types and 1 damage scenario) [12] . The three damage scenarios of NTCG are based on the researches of city of Volos by Kappos et al. [22] (2002), by ITSAK-AUTH (2004) [23] and ARISTION project [17] . The NTCG vulnerability curves have been derived from a hybrid approach [14] [22] , which combines statistical data with appropriately processed (utilising repair cost models) results from nonlinear dynamic or static analyses, that permit extrapolation of statistical data to PGA’s and/or spectral displacements for which no data are available. On the other hand, the pre-mentioned DPMs have been obtained from the empirical (or statistical) seismic method of vulnerability analysis based on processing of a large damage database (which has been cre- ated in the RC laboratory of DUTh [21] ) after the elaboration of the results from post-earthquake surveys carried out after the 7^th of September 1999 Parnitha’s earthquake. The database comprises 180,945 buildings which developed damage of varying degree, type and extent [24] . The damage calibration of the damage dataset was initially based on instructions provided by Earthquake Planning and Protection Organization of Greece (EPPO) and referred to the qualitative characterization for the recording of damage in post-earthquake surveys in Greece [25] [26] . In a recently proposed damage scale a measurable calibration of seismic damage has been presented ac- cording to the physical description and, as well, in terms of structural and economic damage index [27] . Com- paring the total number of damaged buildings to the total number of buildings in the affected area it is concluded that the dataset addresses the 24% of the total number of buildings in the studied area, which is a very wide and reliable statistical sample for buildings. In the collected data, there was no information about the repair costs or the physical description of damage.

The Median Damage Factors (%) for the four different damage scenarios MDFij (%):,(categorized in 7 structural types with 3 damage scenarios: Volos, ΙTSAK-AUTH and ΑRIS- TION, [1] -ΜDFi4 (%), , (categorized in 5 structural types with 1 damage scenario) [12] [20] [21] ) and the seismic demand characterized by the ratio, for several building types are presented in Table 4. A comparison analysis for a specific building type (Figure 2) is conducted according to the existing vulnerability curves.

Parnitha’s near field earthquake on the 7^th of September, 1999 occurred at a small epicentral dis- tance ( 18 km ) from the historical centre of the city of Athens in Greece, a densely populated area and it is con- sidered the biggest recent natural disaster in Greece regarding the monetary loss. The parameter that characte- rizes the seismic input, in National Technical Chamber of Greece [1] models, has been the ratio, where is the regional PGA which has been evaluated from the estimated in earlier research macroseismic intensi- ties and is the unique value is the PGA that characterizes each municipality of Attica according to the haz- ard map of the 2003 Seismic Code [12] [20] .

The examined building stock (750,085 buildings) refers to 122 regions of Attica. Among them, 80 belong in seismic zone I according to the Greek Seismic Code with equivalent ground acceleration

and 42 are classified in seismic zone II with

. The intensity values that were estimated vary from III to IX regarding the 122 regions according to: 1) the Geodynamic Institute of the National Observatory of Athens; 2) a research programme; and 3) isoseismal maps. The majority of the regions belong to weak intensity level and only a few municipalities are found in the area encircled by high intensity isoseismals. The assumption that each municipal- ity has a certain level of seismic severity was necessary for the development of Damage Probability Matrices (DPMs). PGA’s and the corresponding ratios have been evaluated, as they are presented in Table 5. It is important to mention that, beyond the above procedure, an additional loss scenario for the numerical value of

Estimated macroseismic intensity values according to the Modified Mercalli Scale: 1) Geodynamic Institute of the National Observatory of Athens [28] (Kalogeras and Stavrakakis, 2001); 2) Research programme referring to the meizoseismal area [29] (Gazetas and collaborators, 2001); 3) Isoseismal in- tensity maps ([30] Schenková et al. 2007, [31] Hutchings, et al., 2007).

the ratio has also been examined.

According to the National Technical Chamber of Greece [1] , for buildings designed and constructed accord- ing the 1^st Seismic Code of 1959 standing up to 1985 or the 1^st Seismic Code of 1959 plus the 1985 Supplement Clauses (1985-1995), with different from today’s design seismic code, a relative coefficient has been used in order to take into consideration the change in the applied codes. For those buildings that belong in regions, that the design-year seismic zone identification and PGA differs from today’s (Seismic Code 2003) seismic zone and seismic design acceleration, a relative coefficient is used in order to take into account the change by overestimating the Median Damage Factor. A coefficient factor of Median Damage Factor has been estimated according to NPEMEB expressions (Equation (1) and Equation (2)). Table 6 presents the modified design PGAs in several seismic hazard zones according to different Greek seismic codes (SC) and the evaluated modification damage factor.

Summarizing, two different scenarios for soil conditions (a = good soil conditions-smaller PGAs and b = me- dium soil conditions-bigger PGAs) and four damage scenarios have been applied in the applied methodology for seismic risk estimation. An alternative scenario for has also been examined. Moreover, two different scenarios for seismic demand expressed by the ratio have been examined taking into account the eva- luated factor and for, as it has been already discussed. The results of the seismic risk assessment are presented in a map of the study region. Note that in the cases that the coefficient factor has been taken into account by multiplying with for the Median Damage Factor resulted in exaggerated values (over

100%) the upper limit of the vulnerability curve (MDF for) has been adopted. These cases referred in meizoseismal areas with values equal to 1.63, 2.44, 3.42 regarding the most vulnerable buildings.

A pilot methodology is presented herein for the seismic loss assessment in monetary terms in Attica according to the National Programme for Earthquake Management of Existing Buildings [1] [32] . The building stock of At- tica (750,085 buildings) collected from NSSG has been categorized in 7 structural types for the 3 damage sce- narios of NTCG and in 5 structural types for the 4^th damage scenario (Table 4).

The seismic loss factor (in monetary terms) is calculated according to the economic Mean Damage Factor % for each building type () by evaluating the mean ratio of repair/strengthening or replacement cost to the replacement cost of the building with the application of Equation (3). Therefore the re- placement cost of each building is evaluated by the total area and the compatible replacement cost per unit area (?m²).

: total number of buildings belonging to the building type;

: repair/strengthening or replacement cost of the building (?;

: replacement cost of the building (?;

: total area of the building (m²);

: compatible replacement cost per unit area (?m²).

The seismic loss factors, and therefore the estimation of seismic risk, are calculated for every structural type regarding the entire studied area of Attica. The seismic risk loss factors for the four damage scenarios, , and are defined according to the Equation (4) and Equation (5). The mean value of the pre-men- tioned indices is evaluated for the three damage scenarios derived from the NTCG vulnerability models (Equa- tion (6)) and it is compared to the numerical value based on the recently developed DPMs after the Par- nitha’s earthquake.

The normalized seismic risk ratio (%) regarding the total number of buildings of entire Attica is estimated from the mean value divided to the total area of the buildings situated in Attica, as it is presented in Equation (7). The seismic risk ratio regarding the total number of buildings in Greece, , is estimated according to Equation (8) from the numerical value R₄ divided to the total area of the building stock, respectively. It is considered that the building exposure of Greece (Year 2000) refers to 3,990,970 buildings with 6,635,860 floors and estimated mean area per floor 100 m ² . Finally, the seismic loss estimation (in monetary terms) is estimated from the replacement cost (? of the buildings derived from the application of Equation (9).

It is important to clarify that the estimated monetary loss does not include indirect loss (casualties, injuries, interruption of jobs etc.).

Information regarding the compatible repair cost after the 1999 Athens earthquake has also been collected. The statistically derived repair cost for the area is compared with the results of the economic loss estimation ob- tained using the pre-described procedure for the risk assessment. The comparison of the estimated with the compatible cost calibrates the reliability of the commonly used method for the risk assessment and would serve in subsequent earthquake loss estimation studies. The reliable seismic risk management is of crucial importance for the improvement of seismic security and sets the criteria for prioritizing seismic rehabilitation programmes for existing buildings.

The application of the aforementioned methodology requires the distribution of Attica building stock (750,085 buildings) selected from the National Statistics Service of Greece according to the statistical census 2000 in dis- tinct severity levels of seismic input, expressed by the ratio, as it has been already explained (Figure 3). Beyond that, the classification of buildings in structural types together with the total area regarding the building category in each level of ground motion constitutes an essential step for seismic risk assessment (Figure 4).

The results of seismic risk assessment are presented in Table 7 and Figure 5 for the entire examined area of attica including 750,085 buildings, equivalent to 222,748,853 m², for all different damage scenarios that have been above explained. In Figure 6 is presented the distribution of Attica buildings categorized in structural types according to number of buildings, total area and the estimated seismic risk for soil type b.

Note that, the inclusion of the coefficient parameter overestimates significantly the seismic losses. More- rover, the results of the 1^st (Volos) and 2^nd (ITSAK-AUTH) damage scenarios are close, the 3^rd (ARISTION) differs overestimating seismic risk while the 4^th [21] scenario presents the lower values due to the fact that the vulnerability models have been derived from the actual response of the exposed building stock to the referring earthquake. Figure 7 presents all regions of Attica categorized in the severity levels of the seismic input expressed in as they have been estimated from Parnitha’s earthquake and the hazard seismic zones (SC 2003). The estimated seismic risk based on the above methodology is also demonstrated in the same figure.

The statistically derived compatible repair/strengthening or replacement cost has been calculated for the affected area and afterwards it is compared with the results of the economic loss estimation obtained from the application of the pre-described methodology for the risk assessment. It is important to clarify that the estimated monetary loss does not include indirect losses (casualties, injuries, loss of machines/furniture, stop of functions, etc.). The analytical estimation of the statistical repair cost needed the discrimination of damaged buildings from

Parnitha’s earthquake in groups per damage level. Damage data include 178,578 buildings (Table 8) of the cre- ated database and represents the largest existing database in Greece [12] [21] . Moreover, it derived from the same seismic event (7^th-9-1999 Parnitha’s earthquake) with the one used for the simulation of ground motion in the methodology for the predicted losses.

The statistical compatible repair cost was based on two previous researches regarding damaged buildings after the 7^th-9-1999 Parnitha’s earthquake in the region of 1) Aharnes [6] and 2) Ano Liosia (in similar form) [7] be- longing in the epicentral area where heavy damages were recorded. In addition, the mean values of the above researches are also examined (Table 12). The total statistical compatible repair cost of the 178,578 buildings has been evaluated from the mean repair cost per square meter and the mean constructed area per building for each damage category, provided by the Departments for Seismic Restoration in the above mentioned researches, as it is presented in Tables 9-11.

According to research (1) of Aharnes the damage data included 2556 buildings with 640,174 (m²) total area and total repair cost 80.50 (M?. Among these buildings: 1) 51 developed minor (green) damages with 12,610 (m²) total area and 0.41 (M? approved total repair cost; 2) 1586 developed moderate (yellow) damages with 452,658 (m²) total area and approved total repair cost 28.19 (M?; 3) 919 developed extensive (red) damages with 174,906 (m²) total area and 51.90 (M? approved total repair cost. According to research (2) of Ano Liosia the damage data included 983 buildings with total area repair 147,392 (m²) and total repair cost 17.21 (M?. Among these buildings: a) 403 developed minor (green) damages with 59,547 (m²) total area and approved total repair cost 2.11 (M?; b) 350 developed moderate (yellow) damages with 61,871 (m²) total area and approved total repair cost 5.726 (M?; c) 230 developed extensive (red) damages with 25,974 (m²) total area and approved total repair cost 9.38 (M?.

Based on the statistical data of Aharnes the total compatible repair cost has been evaluated equal to 2419.71 M? with equivalent replacement area of buildings 8.15 km² (Table 9). According to the statistical data of Ano Liosia the total compatible repair cost has been evaluated equal to 1869.72 M?with equivalent replacement area 5.18 km² (Table 10). Finally, taking the mean values for the repair cost of the two researches including 3539 buildings the total compatible repair cost has been evaluated equal to 2095.22 M?(Table 11) with equivalent replacement area of buildings 6.87 km².

The statistical costs for each damage category are presented in Table 12. It has resulted that the mean statisti- cal compatible replacement cost has risen in 297 ?m², 361 ?m² or 305 ?m², taking into consideration the two researches and the mean value, respectively. The last values have been adopted for the evaluation of the equi- valent replacement area (m²) from the total repair/strengthening cost per discrete damage levels in Tables 9-11. The evaluation and distribution of the statistical repair cost per intensity and damage level is presented in Table 13 for the 178,578 damaged buildings based on the statistical data of the two researches.

Note that the approved budget by the National Agency for the Relief of Earthquake Victims for repair/replace- ment cost was based on a compatible work invoice. One third (1/3) of the approved budget was national assis- tance and the rest (2/3) was provided as without interest loan in earthquake victims. The upper limits that were set by the National Agency for the Relief of Earthquake Victims for the replacement and repair cost for habitat use regarding a building with area up to 120 m² were 382 ?m² and 191 ?m², respectively. The mean actual re- placement cost in the same period was 528 ?m² according to estimations of the National Technical Chamber of Greece. Thus the reduction of the compatible to the actual cost could be attained by multiplying the first with the values of 1.78 (528/297 ?m²), 1.46 (528/361 ?m²) and 1.73 (528/305 ?m²), regarding the 1) Aharnes research, 2) Ano Liosia and both (1 + 2) researches, respectively.

Finally, the created damage database referred in 180,945 buildings. Among them the 180,427 had the charac- terization of damage and the 178,578 were also able to be discriminated in structural types. Following the same assumptions an additional assessment of the compatible and the actual repair cost was fulfilled by multiplying with the values of 1.78, 1.46 and 1.73, as it is presented in Tables 14-16, regarding the Aharnes research, Ano Liosia and both (1 + 2) researches, respectively.

The results of seismic risk assessment are presented in Table 17 for the entire examined area of Attica including 750,085 buildings for four different damage scenarios along with the estimated compatible cost in monetary loss based on the above mentioned. Conducting a comparison analysis between the predicted with the statistical

compatible cost it is concluded that generally the seismic risk methodology overestimates seismic losses. As expected, the seismic scenario based on the developed DPMs [21] from 7^th-9-1999 Athens damage data pre- sented the better correlation (2627.77 M? with the total statistically evaluated repair cost, especially when the last was based on Aharnes research (2450.02 M?. It is important to stress that the inclusion of the coefficient parameter overestimates significantly the seismic losses. The last result should be taken into consideration in future risk researches.

A complete research of seismic risk assessment is presented regarding the extended urban region of Athens in Greece. The seismic risk assessment is fulfilled by discriminating the current study in two approaches, probable and actual, conducting afterwards between them a comparison analysis. In the first part, a pilot methodology is developed for the seismic loss assessment in monetary terms regarding the buildings damages, consistent with the National Programme for Earthquake Management of Existing Buildings (NPEMEB). The building stock consists of typical building types of Southern Europe and refers to 750,085 buildings (18.80% of buildings in Greece) situated in the entire region of Athens according to the results of the 2000-1 statistical census. A wider research of seismic risk assessment could include direct losses of infrastructures and indirect economic losses. The evaluation of loss due to building damage in a certain region requires an assessment of both seismic hazard and vulnerability of the building stock in the study area. Three different scenarios for soil conditions (a, b and) and four damage scenarios have been applied in the described methodology for the estimation of the seismic risk. The results of the seismic risk assessment for the four different aspects of the estimated damage and the different soil conditions have been presented in a map of the study region. The existing vulnerability curves corresponding to defined types of buildings have been derived from the National Technical Chamber of Greece and also from recently developed DPMs. The last DPMs were obtained in a previous research [21] from the process of a created damage database after the 7^th of September 1999 Parnitha’s earthquake and comprised 180,945 buildings which developed damage of varying degree, type and extent. The numerical values of the seismic risk factors (Table 7 & Table 17) after the application of the described methodology are estimated as follows:

1) , and with predicted repair cost from 2627.77 to 3194.02 M? for according to the Parnitha’s earthquake and.

2) , and, for and.

3) , and with predicted repair cost from 7300.92 to 8874.18 M? for according to the Parnitha’s earthquake and for soil type.

4) , and with predicted repair cost from 4800.83 to 5835.35 M? for according to the Parnitha’s earthquake and for soil type.

5) , and, for and for soil type.

6) , and, for and for soil type.

In the second part of the research, the seismic risk is evaluated from the available data regarding the mean sta- tistical repair/strengthening or replacement cost for the total number of damaged structures (180,427 buildings) after the same (1999 Parnitha’s) seismic event. Data regarding the compatible (budget approved according to the ministry’s provisions) repair cost has been collected. The structural losses in monetary terms for the 180,427 buildings damaged structures are evaluated equal to 2450.0 Μ? 1887.8 Μ?and 2118.9 Μ?based on the pre- viously mentioned statistical seismic risk data. The statistically derived repair cost for Attica is compared with the results of the economic loss estimation for buildings using the aforementioned risk assessment methodology. Conducting a comparison analysis between the estimated with the compatible repair cost it is concluded that generally the seismic risk methodology overestimates seismic losses. It should be mentioned, though, that the predicted loss takes into consideration the total building stock and not only the damaged buildings. From the analysis results, the seismic scenario based on the recently developed DPMs [21] presented the better correlation (2627.77 M? with the total statistically evaluated repair cost (2450.02 M?. It is important to stress that the in- clusion of the coefficient parameter S overestimates significantly the seismic losses. The last result should be ta- ken into consideration in future risk researches. The benefits which arise from the research are connected to in- dividuals, engineers and citizens, and also governments, research centres or organizations related to the earth- quake management and protection. The comparison of the estimated economic loss with the actual repair cost calibrates the reliability of the commonly used method for the risk assessment and serves in the improvement of seismic security and prioritizing the criteria for seismic rehabilitation programmes of existing buildings.